X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=ids-find.el;h=eb404b356120a895f36e8ce2b53f9cdf86d1ddc4;hb=f641879e4d25e23e8c63085a64cf5c83390f8737;hp=ad9ff9c4e947ab3cb1889cf52de8c2fc2e364590;hpb=a470b505b5753f9d1951923b233e31be13a9a467;p=chise%2Fids.git

diff --git a/ids-find.el b/ids-find.el
index ad9ff9c..eb404b3 100644
--- a/ids-find.el
+++ b/ids-find.el
@@ -1,11 +1,11 @@
 ;;; ids-find.el --- search utility based on Ideographic-structures
 
-;; Copyright (C) 2002 MORIOKA Tomohiko
+;; Copyright (C) 2002,2003,2005,2006,2007,2017 MORIOKA Tomohiko
 
 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
-;; Keywords: Kanji, Ideographs, search, IDS
+;; Keywords: Kanji, Ideographs, search, IDS, CHISE, UCS, Unicode
 
-;; This file is a part of Tomoyo-Tools.
+;; This file is a part of CHISE IDS.
 
 ;; This program is free software; you can redistribute it and/or
 ;; modify it under the terms of the GNU General Public License as
@@ -24,6 +24,186 @@
 
 ;;; Code:
 
+(defun ids-index-store-char (product component)
+  (let ((ret (get-char-attribute component 'ideographic-products)))
+    (unless (memq product ret)
+      (put-char-attribute component 'ideographic-products
+			  (cons product ret))
+      (when (setq ret (char-feature component 'ideographic-structure))
+	(ids-index-store-structure product ret)))
+    ))
+
+(defun ids-index-store-structure (product structure)
+  (let (ret)
+    (dolist (cell (cdr structure))
+      (if (char-ref-p cell)
+	  (setq cell (plist-get cell :char)))
+      (cond ((characterp cell)
+	     (ids-index-store-char product cell))
+	    ((setq ret (assq 'ideographic-structure cell))
+	     (ids-index-store-structure product (cdr ret)))
+	    ((setq ret (find-char cell))
+	     (ids-index-store-char product ret))
+	    ))))
+
+;;;###autoload
+(defun ids-update-index ()
+  (interactive)
+  (map-char-attribute
+   (lambda (c v)
+     (ids-index-store-structure c v)
+     nil)
+   'ideographic-structure)
+  (save-char-attribute-table 'ideographic-products))
+
+
+(mount-char-attribute-table 'ideographic-products)
+
+;;;###autoload
+(defun ids-find-all-products (char)
+  (let (dest)
+    (dolist (cell (char-feature char 'ideographic-products))
+      (unless (memq cell dest)
+	(setq dest (cons cell dest)))
+      (setq dest (union dest (ids-find-all-products cell))))
+    dest))
+
+(defun of-component-features ()
+  (let (dest)
+    (dolist (feature (char-attribute-list))
+      (when (string-match "^<-.*[@/]component\\(/[^*/]+\\)*$"
+			  (symbol-name feature))
+	(push feature dest)))
+    (list* '<-mistakable '->mistakable
+	   '<-formed '->formed
+	   '<-same '->same
+	   '<-original '->original
+	   '<-ancient '->ancient
+	   dest)))
+
+(defun to-component-features ()
+  (let (dest)
+    (dolist (feature (char-attribute-list))
+      (when (string-match "^->.*[@/]component\\(/[^*/]+\\)*$"
+			  (symbol-name feature))
+	(push feature dest)))
+    dest))
+
+;;;###autoload
+(defun char-component-variants (char)
+  (let ((dest (list char))
+	ret uchr)
+    (dolist (feature (to-component-features))
+      (if (setq ret (get-char-attribute char feature))
+	  (dolist (c ret)
+	    (setq dest (union dest (char-component-variants c))))))
+    (cond
+     ;; ((setq ret (some (lambda (feature)
+     ;;                    (get-char-attribute char feature))
+     ;;                  (to-component-features)))
+     ;;  (dolist (c ret)
+     ;;    (setq dest (union dest (char-component-variants c))))
+     ;;  )
+     ((setq ret (get-char-attribute char '->ucs-unified))
+      (setq dest (cons char ret))
+      (dolist (c dest)
+	(setq dest (union dest
+                          (some (lambda (feature)
+				  (get-char-attribute c feature))
+				(of-component-features))
+			  )))
+      )
+     ((and (setq ret (get-char-attribute char '=>ucs))
+	   (setq uchr (decode-char '=ucs ret)))
+      (setq dest (cons uchr (char-variants uchr)))
+      (dolist (c dest)
+	(setq dest (union dest
+                          (some (lambda (feature)
+				  (get-char-attribute c feature))
+				(of-component-features))
+			  )))
+      )
+     (t
+      (map-char-family
+       (lambda (c)
+	 (unless (memq c dest)
+	   (setq dest (cons c dest)))
+	 (setq dest
+	       (union dest
+		      (some (lambda (feature)
+			      (char-feature c feature))
+			    (of-component-features))
+		      ))
+	 nil)
+       char)
+      ))
+    dest))
+
+;;;###autoload
+(defun ideographic-products-find (&rest components)
+  (if (stringp (car components))
+      (setq components (string-to-char-list (car components))))
+  (let (dest products)
+    (dolist (variant (char-component-variants (car components)))
+      (setq products
+	    (union products
+		   (get-char-attribute variant 'ideographic-products))))
+    (setq dest products)
+    (while (and dest
+		(setq components (cdr components)))
+      (setq products nil)
+      (dolist (variant (char-component-variants (car components)))
+	(setq products
+	      (union products
+		     (get-char-attribute variant 'ideographic-products))))
+      (setq dest (intersection dest products)))
+    dest))
+
+(defun ideograph-find-products-with-variants (components &optional ignored-chars)
+  (if (stringp components)
+      (setq components (string-to-char-list components)))
+  (let (dest products)
+    (dolist (variant (char-component-variants (car components)))
+      (setq products
+	    (union products
+		   (set-difference
+		    (get-char-attribute variant 'ideographic-products)
+		    ignored-chars))))
+    (setq dest products)
+    (while (and dest
+		(setq components (cdr components)))
+      (setq products nil)
+      (dolist (variant (char-component-variants (car components)))
+	(setq products
+	      (union products
+		     (set-difference
+		      (get-char-attribute variant 'ideographic-products)
+		      ignored-chars))))
+      (setq dest (intersection dest products)))
+    dest))
+
+(defun ideograph-find-products (components &optional ignored-chars)
+  (if (stringp components)
+      (setq components (string-to-char-list components)))
+  (let (dest products)
+    ;; (dolist (variant (char-component-variants (car components)))
+    ;;   (setq products
+    ;;         (union products
+    ;;                (get-char-attribute variant 'ideographic-products))))
+    ;; (setq dest products)
+    (setq dest (get-char-attribute (car components) 'ideographic-products))
+    (while (and dest
+		(setq components (cdr components)))
+      ;; (setq products nil)
+      ;; (dolist (variant (char-component-variants (car components)))
+      ;;   (setq products
+      ;;         (union products
+      ;;                (get-char-attribute variant 'ideographic-products))))
+      (setq products (get-char-attribute (car components) 'ideographic-products))
+      (setq dest (intersection dest products)))
+    dest))
+
+
 (defun ideographic-structure-char= (c1 c2)
   (or (eq c1 c2)
       (and c1 c2
@@ -31,88 +211,200 @@
 		 (m2 (char-ucs c2)))
 	     (or (and m1 m2
 		      (eq m1 m2))
-		 (progn
-		   (setq m1 (car (get-char-attribute c1 '<-radical))
-			 m2 (car (get-char-attribute c2 '<-radical)))
-		   (unless (characterp m1)
-		     (setq m1 (or (find-char m1))))
-		   (unless (characterp m2)
-		     (setq m2 (find-char m2)))
-		   (when (or m1 m2)
-		     (ideographic-structure-char= m1 m2))))))))
-
-(defun ideographic-structure-member-compare-parts (part s-part)
+		 (memq c1 (char-component-variants c2)))))))
+
+(defun ideographic-structure-member-compare-components (component s-component)
   (let (ret)
-    (cond ((char-ref= part s-part #'ideographic-structure-char=))
-	  ((listp s-part)
-	   (if (setq ret (assq 'ideographic-structure s-part))
-	       (ideographic-structure-member part (cdr ret))))
-	  ((setq ret (get-char-attribute s-part 'ideographic-structure))
-	   (ideographic-structure-member part ret)))))
+    (cond ((char-ref= component s-component #'ideographic-structure-char=))
+	  ((listp s-component)
+	   (if (setq ret (assq 'ideographic-structure s-component))
+	       (ideographic-structure-member component (cdr ret))))
+	  ((setq ret (get-char-attribute s-component 'ideographic-structure))
+	   (ideographic-structure-member component ret)))))
 
 ;;;###autoload
-(defun ideographic-structure-member (part structure)
-  "Return non-nil if PART is included in STRUCTURE."
-  (or (progn
+(defun ideographic-structure-member (component structure)
+  "Return non-nil if COMPONENT is included in STRUCTURE."
+  (or (memq component structure)
+      (progn
 	(setq structure (cdr structure))
-	(ideographic-structure-member-compare-parts part (car structure)))
+	(ideographic-structure-member-compare-components
+	 component (car structure)))
       (progn
 	(setq structure (cdr structure))
-	(ideographic-structure-member-compare-parts part (car structure)))
+	(ideographic-structure-member-compare-components
+	 component (car structure)))
       (progn
 	(setq structure (cdr structure))
 	(and (car structure)
-	     (ideographic-structure-member-compare-parts
-	      part (car structure))))))
+	     (ideographic-structure-member-compare-components
+	      component (car structure))))))
 
 
 ;;;###autoload
-(defun ideographic-structure-repertoire-p (structure parts)
-  "Return non-nil if STRUCTURE can be constructed by a subset of PARTS."
+(defun ideographic-structure-repertoire-p (structure components)
+  "Return non-nil if STRUCTURE can be constructed by a subset of COMPONENTS."
   (and structure
-       (let (ret s-part)
+       (let (ret s-component)
 	 (catch 'tag
 	   (while (setq structure (cdr structure))
-	     (setq s-part (car structure))
-	     (unless (characterp s-part)
-	       (if (setq ret (find-char s-part))
-		   (setq s-part ret)))
+	     (setq s-component (car structure))
+	     (unless (characterp s-component)
+	       (if (setq ret (find-char s-component))
+		   (setq s-component ret)))
 	     (unless (cond
-		      ((listp s-part)
-		       (if (setq ret (assq 'ideographic-structure s-part))
+		      ((listp s-component)
+		       (if (setq ret (assq 'ideographic-structure s-component))
 			   (ideographic-structure-repertoire-p
-			    (cdr ret) parts)))
-		      ((member* s-part parts
+			    (cdr ret) components)))
+		      ((member* s-component components
 				:test #'ideographic-structure-char=))
 		      ((setq ret
-			     (get-char-attribute s-part
+			     (get-char-attribute s-component
 						 'ideographic-structure))
-		       (ideographic-structure-repertoire-p ret parts)))
+		       (ideographic-structure-repertoire-p ret components)))
 	       (throw 'tag nil)))
 	   t))))
 
+
+(defvar ids-find-result-buffer "*ids-chars*")
+
+(defun ids-find-format-line (c v)
+  (format "%c\t%s\t%s\n"
+	  c
+	  (or (let ((ucs (or (char-ucs c)
+			     (encode-char c 'ucs))))
+		(if ucs
+		    (cond ((<= ucs #xFFFF)
+			   (format "    U+%04X" ucs))
+			  ((<= ucs #x10FFFF)
+			   (format "U-%08X" ucs)))))
+	      "          ")
+	  (or (ideographic-structure-to-ids v)
+	      v)))
+
+(defun ids-insert-chars-including-components* (components
+					       &optional level ignored-chars)
+  (unless level
+    (setq level 0))
+  (let (is i as bs)
+    (dolist (c (sort (copy-tree (ideograph-find-products components
+							 ignored-chars))
+		     (lambda (a b)
+		       (if (setq as (char-total-strokes a))
+			   (if (setq bs (char-total-strokes b))
+			       (if (= as bs)
+				   (ideograph-char< a b)
+				 (< as bs))
+			     t)
+			 (ideograph-char< a b)))))
+      (unless (memq c ignored-chars)
+	(setq is (char-feature c 'ideographic-structure))
+	(setq i 0)
+	(while (< i level)
+	  (insert "\t")
+	  (setq i (1+ i)))
+	(insert (ids-find-format-line c is))
+	(setq ignored-chars
+	      (ids-insert-chars-including-components*
+	       (char-to-string c) (1+ level)
+	       (cons c ignored-chars))))
+      )
+    )
+  ignored-chars)
+
+(defun ids-insert-chars-including-components (components
+					      &optional level ignored-chars)
+  (unless level
+    (setq level 0))
+  (setq ignored-chars
+	(nreverse
+	 (ids-insert-chars-including-components* components
+						 level ignored-chars)))
+  (let (is i as bs)
+    (dolist (c ignored-chars)
+      (dolist (vc (char-component-variants c))
+	(unless (memq vc ignored-chars)
+	  (when (setq is (get-char-attribute vc 'ideographic-structure))
+	    (setq i 0)
+	    (while (< i level)
+	      (insert "\t")
+	      (setq i (1+ i)))
+	    (insert (ids-find-format-line vc is))
+	    (setq ignored-chars
+		  (ids-insert-chars-including-components*
+		   (char-to-string vc) (1+ level)
+		   (cons vc ignored-chars)))))))
+    (dolist (c (sort (copy-tree (ideograph-find-products-with-variants
+				 components ignored-chars))
+		     (lambda (a b)
+		       (if (setq as (char-total-strokes a))
+			   (if (setq bs (char-total-strokes b))
+			       (if (= as bs)
+				   (ideograph-char< a b)
+				 (< as bs))
+			     t)
+			 (ideograph-char< a b)))))
+      (unless (memq c ignored-chars)
+	(setq is (get-char-attribute c 'ideographic-structure))
+	(setq i 0)
+	(while (< i level)
+	  (insert "\t")
+	  (setq i (1+ i)))
+	(insert (ids-find-format-line c is))
+	(setq ignored-chars
+	      (ids-insert-chars-including-components*
+	       (char-to-string c) (1+ level)
+	       (cons c ignored-chars))))
+      )
+    )
+  ignored-chars)
+
+;;;###autoload
+(defun ids-find-chars-including-components (components)
+  "Search Ideographs whose structures have COMPONENTS."
+  (interactive "sComponents : ")
+  (with-current-buffer (get-buffer-create ids-find-result-buffer)
+    (setq buffer-read-only nil)
+    (erase-buffer)
+    (ids-insert-chars-including-components components 0 nil)
+    ;; (let ((ignored-chars
+    ;;        (nreverse
+    ;;         (ids-insert-chars-including-components components 0 nil
+    ;;                                                #'ideograph-find-products)))
+    ;;       rest)
+    ;;   (setq rest ignored-chars)
+    ;;   ;; (dolist (c rest)
+    ;;   ;;   (setq ignored-chars
+    ;;   ;;         (union ignored-chars
+    ;;   ;;                (ids-insert-chars-including-components
+    ;;   ;;                 (list c) 0 ignored-chars
+    ;;   ;;                 #'ideograph-find-products-with-variants))))
+    ;;   (ids-insert-chars-including-components components 0 ignored-chars
+    ;;                                          #'ideograph-find-products-with-variants))
+    (goto-char (point-min)))
+  (view-buffer ids-find-result-buffer))
+
+;;;###autoload
+(define-obsolete-function-alias 'ideographic-structure-search-chars
+  'ids-find-chars-including-components)
+
 ;;;###autoload
-(defun ideographic-structure-search-chars (parts)
-  "Search Ideographs by PARTS."
-  (interactive "sParts : ")
-  (with-current-buffer (get-buffer-create " *ids-chars*")
+(defun ids-find-chars-covered-by-components (components)
+  "Search Ideographs which structures are consisted by subsets of COMPONENTS."
+  (interactive "sComponents: ")
+  (if (stringp components)
+      (setq components (string-to-char-list components)))
+  (with-current-buffer (get-buffer-create ids-find-result-buffer)
     (setq buffer-read-only nil)
     (erase-buffer)
     (map-char-attribute
      (lambda (c v)
-       (when (every
-	      (lambda (p)
-		;; (member* p v :test #'char-ref=)
-		(ideographic-structure-member p v))
-	      parts)
-	 (insert (format "%c\t%s\n"
-			 c
-			 (or (ideographic-structure-to-ids v)
-			     v))))
-       nil)
+       (when (ideographic-structure-repertoire-p v components)
+	 (insert (ids-find-format-line c v))))
      'ideographic-structure)
     (goto-char (point-min)))
-  (view-buffer " *ids-chars*"))
+  (view-buffer ids-find-result-buffer))
 
 
 ;;; @ End.